Data for: "Proof-of-concept trial of electronic monitoring devices for adolescents living with HIV in Zimbabwe" – Data Codebook

Persistent identifier

10.17037/DATA.00005346

Description

The VITALITY randomised controlled trial investigated the effectiveness of vitamin D and calcium supplementation on bone density in adolescents living with HIV. Nested within VITALITY, this pilot trial investigated the effectiveness of electronic monitoring devices (EMDs) on adherence to vitamin D supplementation in Harare, Zimbabwe.

169 adolescents living with HIV were randomised to receive an EMD (N=97) or no EMD (n=72) for 24 weeks. 84 of the adolescents were taking vitamin D supplements (N=50 with EMD, N=34 without EMD). The pilot trial compared 25OHD levels in blood between the EMD and non-EMD participants receiving vitamin D supplements after 24 weeks as a marker of adherence.

This record contains two datasets, an R script, Stata do file, and accompanying documentation:

1. The ‘EMD_minimal_deidentified’ dataset contains records on 97 participants who were randomly allocated to receive an electronic monitoring device and 72 participants who were randomly allocated to not receive one. It can be analysed with the Stata do file “Minimal dataset.do”, which replicates the pilot trial analysis of adherence and vitamin D level by EMD arm, among those who were receiving vitamin D supplement in the parent VITALITY trial.
2. The ‘ml_preprocessed_for_R’ dataset contains records on 97 participants who received an electronic monitoring device. It can be analysed with the R script “clustering_deidentified.R, which reproduces the unsupervised machine learning techniques for exploratory analysis of patterns of EMD use over 24 weeks, using k-means clustering, k-medoids clustering, hierarchical agglomerative clustering and Gaussian mixture model.

The original study's ethical approvals state data will be used in ethically approved research only. Data will be shared once evidence of ethics approval is provided.

Data Codebook

This record contains two datasets and accompanying scripts to analyse them:

• EMD_minimal_deidentified
• ml_preprocessed_for_R

EMD_minimal_deidentified

The ‘EMD_minimal_deidentified’ dataset contains records on 97 participants who were randomly allocated to receive an electronic monitoring device and 72 participants who were randomly allocated to not receive one. It can be analysed with the Stata do file “Minimal dataset.do”.  This do file replicates the pilot trial analysis of adherence and vitamin D level by EMD arm, among those who were receiving vitamin D supplement in the parent VITALITY trial.

The original study's ethical approvals state data will be used in ethically approved research only. Data will be shared once evidence of ethics approval is provided.

ml_preprocessed_for_R

The ‘ml_preprocessed_for_R’ dataset contains records on 97 participants who received an electronic monitoring device. It can be analysed with the R script clustering_deidentified. This R script reproduces the unsupervised machine learning techniques for exploratory analysis of patterns of EMD use over 24 weeks, using k-means clustering, k-medoids clustering, hierarchical agglomerative clustering and Gaussian mixture model.